NEDDylation is essential for Kaposi's sarcoma-associated herpesvirus latency and lytic reactivation and represents a novel anti-KSHV target

Abstract

Kaposi's sarcoma-associated herpesvirus (KSHV) is the causative agent of Kaposi's sarcoma (KS) and primary effusion lymphoma (PEL), which are aggressive malignancies associated with immunocompromised patients. For many non-viral malignancies, therapeutically targeting the ubiquitin proteasome system (UPS) has been successful. Likewise, laboratory studies have demonstrated that inhibition of the UPS might provide a promising avenue for the treatment of KSHV-associated diseases. The largest class of E3 ubiquitin ligases are the cullin-RING ligases (CRLs) that are activated by an additional ubiquitin-like protein, NEDD8. We show that pharmacological inhibition of NEDDylation (using the small molecule inhibitor MLN4924) is cytotoxic to PEL cells by inhibiting NF-κB. We also show that CRL4B is a novel regulator of latency as its inhibition reactivated lytic gene expression. Furthermore, we uncovered a requirement for NEDDylation during the reactivation of the KSHV lytic cycle. Intriguingly, inhibition prevented viral DNA replication but not lytic cycle-associated gene expression, highlighting a novel mechanism that uncouples these two features of KSHV biology. Mechanistically, we show that MLN4924 treatment precluded the recruitment of the viral pre-replication complex to the origin of lytic DNA replication (OriLyt). These new findings have revealed novel mechanisms that regulate KSHV latency and reactivation. Moreover, they demonstrate that inhibition of NEDDylation represents a novel approach for the treatment of KSHV-associated malignancies.

Fig 1. Inhibition of NEDDylation is cytotoxic in PEL cells.

(A) TREx-BCBL-1-RTA and BC-3 cells were plated into 96-well plates and treated at varying concentrations of MLN4924. Cells viability was determined 96 h later using an ATPlite assay, and shown is a representative trace of several experiments, and EC50s were calculated (TREx-BCBL-1-RTA = 1.10 μM; BC-3 = 0.15 μM). (B) Immunoblot analysis of TREx-BCBL-1-RTA cells showing that MLN4924 effectively inhibits NEDDylation in PEL cells as shown by reductions in NEDDylated Cul2 after 24 h treatment. Also, treatment appeared to reactivate KSHV lytic gene expression (see ORF57 expression in 10 μM lane). (C) Similar to (B), but with a second PEL cell line, BC-3. As for TREx-BCBL-1-RTA, treatment led to the activation of lytic gene expression as shown by endogenous RTA expression. Cleavage of PARP also indicated that MLN4924 activates apoptosis in this cell line. (D) Cell cycle analysis of TREx-BCBL-1-RTA cells treated for 24 h with 1 μM MLN4924 showing a reduction in S-phase and an accumulation of cells in G2/M. (E) TREx-BCBL-1-RTA cells treated with 10 μM MLN4924 for 24 h activated caspase 3/7 as indicated by PARP-1 cleavage. In contrast to lytic reactivation by Dox., MLN4924 led to the activation of initiator Caspase 9 which presumably led to Caspase 3 cleavage (activation). (F) Stationary cultures of BCBL-1 cells were treated with MLN4924 and cell viability was monitored by trypan blue exclusion over 5 days. Cells were pre-treated with z-VAD-FMK just prior to treatment to inhibit caspase-mediated apoptosis; however, this did not prevent cytotoxicity. FL PARP-1 denotes full length PARP-1; Cl. PARP-1 denotes cleaved PARP-1. FL ORF57 denotes full length ORF57 and Cl. ORF57 denotes the caspase 7-cleaved ORF57 [29].

Fig 2. MLN4924 induced lytic cycle—associated expression.

(A) Latently infected TREx-BCBL-1-RTA cells were treated with 1 μM MLN4924 and total RNA was isolated from samples harvested 24 h later. The normalized expression (using GAPDH expression) of lytic cycle genes and latency associated (ORF73, ORF71 and K12) genes were analyzed by qRT-PCR and compared to the levels found in DMSO-treated control samples. Three biological replicates were independently analyzed and the error bars represent the standard deviation from the mean. (B) The inhibition of NEDDylation leads to reactivation of lytic cycle protein expression; rKSHV.219 cells were treated with 0.1 μM MLN4924 (a concentration that was tolerated for the duration of the experiment) for 36 h. Red fluorescent protein (RFP) expression was observed in numerous MLN4924-treated cells, indicative of RTA protein expression. RFP expression was observed in a minority DMSO-only treated cells (as expected due to low level spontaneous reactivation). (C) ORF57 expression was also detected by immunoblot analysis of MLN4924-treated rKSHV.219 cells. (D) Inhibition of NEDDylation enhances reactivation. MLN4924-treated rKSHV.219 cells were incubated with or without TPA (able to reactivate KSHV) for 36 h.

Fig 3. PEL cytotoxicity is due to NF-κB inhibition.

(A) TREx-BCBL-1-RTA cells were treated with 5 μM MLN4924 for the indicated times. Cell lysates were analyzed by immunoblot using the indicated antibodies to determine the mode of cytotoxicity. (B) Transcription of NF-κB-regulated latency-associated gene LANA was determined at 2, 4, 6 and 8 h post treatment with 5 μM MLN4924 to investigate the point at which the inhibition of NEDDylation had an effect. (C) MLN4924 treatment led to the stabilization of phosphorylated IκBα (pIκBα[ser32]) in TREx-BCBL-1-RTA cells, IκBα expression is reduced in lytic cells reactivated by the addition of Dox. (D) Similar to (C), pIκBα is stabilized in MLN4924-treated BC-3 cells. (E) Cytoplasmic fractions were enriched (as shown by the lack of nuclear Lamin B and enrichment of GAPDH—compare to nuclear fraction [Nuc.]) from TREx-BCBL-1-RTA cells treated with varying concentrations of MLN4924 which shows the stabilization of the p65/RelA subunit of NF-κB in the cytoplasm. (F) Densitometry analysis using ImageJ of the immunoblot analysis shown in (E).

Fig 4. The role of individual CRLs for the regulation of KSHV latency.

(A) Dominant-negative versions of each Cullin (FLAG-DNCul that inhibits CRL activity [35]) was expressed in latently infected rKSHV.219 cells and ORF57 expression (as a marker of lytic reactivation) was tested by immunoblot analysis 36 h later. Treatment of cells with MLN4924 and TPA/NaB (sodium butyrate) served as positive controls for reactivation. *denotes non-specific bands and error bars represent the standard deviation of the mean of two independent transfection experiments. (B) Densitometry analysis of (A). (C) qRT-PCR analysis of Cul4B mRNA from TREx-BCBL-1-RTA cells transfected with either a scramble control shRNA expression vector (Scr.) or four independent shRNA expression vectors targeted at Cul4B (shCul4B). RNA was harvested four days post transfection. (D) qRT-PCR analysis of viral gene expression in Cul4B-knockdown cells normalized against GAPDH expression. Error bars represent the standard deviation from three independent transfections per condition. OLT denotes OriLyt transcript.

Fig 5. NEDDylation is required for lytic reactivation.

(A) To test if NEDDylation-associated ubiquitylation is a feature of KSHV lytic reactivation, immunofluorescence analysis was used. TREx-BCBL-1-RTA cells were left untreated, or treated with MLN4924 and reactivated by the addition of Dox. Soluble nuclear proteins (i.e. those not associated with chromatin) were removed prior to fixation (see Materials and Methods). Ubiquitylated proteins (using antibodies specific for ubiquitylated proteins and not free ubiquitin) were seen associated with the edges of KSHV replication compartments (inset; RC denotes replication compartment; Ub denotes ubiquitin-modified proteins), as indicated by RTA staining in untreated, but dox. reactivated cells and the level of ubiquitylation was reduced in the presence of 1 μM MLN4924. (B) Cell lysates were prepared from TREx-BCBL-1-RTA cells that were treated with MLN4924 and reactivated with Dox for 24 h. Immunoblot analysis of viral proteins shows that lytic-associated protein expression is inhibited in a dose-dependent manner (C) TREx-BCBL-1-RTA cells were treated at varying concentrations of MLN4924, reactivated with doxycycline and total DNA was isolated 72 h later. Normalized (to cellular GAPDH) viral genome replication was measured using qPCR analysis of three biological replicates; error bars represent standard deviation from the mean.

Fig 6. Inhibiting MLN4924-induced apoptosis does not restore lytic reactivation.

(A) Inhibition of caspase activity (using the pan-caspase inhibitor z-VAD-FMK) restored lytic cycle protein expression in reactivated cells treated with MLN4924. Successful inhibition of caspase activity was monitored by immunoblot analysis of PARP-1 and ORF57 cleavage. FL ORF57 denotes full length ORF57; Cl. ORF57 denotes cleaved ORF57. (B) Despite restoration of protein expression upon caspase inhibition, viral genome replication was still blocked. TREx-BCBL-1-RTA cells were treated as indicated and cells were harvested 72 h later. Total DNA was harvested and qPCR was used to determine KSHV genome copies relative to latent cells (as in Fig. 5C).

Fig 7. MLN4924 prevents the proper organization of replication compartments and the recruitment of RTA to OriLyt.

(A) Replication compartments were observed in TREx-BCBL-1-RTA cells reactivated by Dox. for 16 h by imaging for the incorporation of EdU into replicating KSHV DNA, and its colocalization with RTA (arrows, top panel). Treatment with MLN4924 inhibited the formation of replication compartments (as observed by pan-nuclear staining of RTA, bottom panels) and blocked the replication of viral DNA (as indicated by the absence of EdU in RTA-positive cells). (B) Reactivated cells treated with caspase inhibitor z-VAD-FMK still formed replication compartments, yet z-VAD-FMK was unable to restore KSHV DNA replication in MLN4924 treated cells. (C) Latent TREx-BCBL-1-RTA cells were treated with 1 μM MLN4924 for 18 h and ChIP analysis was performed using antibodies specific for IgG (isotype control), RNA Pol II and Myc-tag (for RTA), and primers specific for the RTA responsive element (RRE) of OriLyt. Treatment induced RNA Pol II recruitment (with a slight increase in RTA) corroborating our transcriptional analyses (see Fig. 2). (D) The same analysis was performed but in cells reactivated with Dox. This analysis showed that MLN4924 treatment blocked the recruitment of RTA onto OriLyt during lytic reactivation. Error bars represent the standard deviation of the mean from two independent ChIP experiments.

Fig 8. Summary.

(A) NF-κB signaling is essential for the survival of PEL as it drives the expression of KSHV latency genes, suppresses lytic cycle-associated genes and promotes antiapoptotoic gene expression [28]. Normally, IκBα inhibits NF-κB by preventing its nuclear translocation. Upon stimulation, or modulation by viral proteins such as vFLIP, IκBα is phosphorylated which signals for the recruitment of the Cul1-containing βTrCP ubiquitin ligase, leading to its polyubiquitylation and degradation [32]. This releases NF-κB allowing it to translocate to the nucleus to activate transcription. This study has shown that MLN4924 blocks the degradation of IκBα thus preventing the nuclear translocation of NF-κB and subsequent PEL cell death. (B) The lytic cycle of KSHV is required for the pathogenesis of KS [7], and therefore the molecular mechanisms governing KSHV reactivation need to be understood. This study has shown that inhibiting NEDDylation prevents lytic reactivation by blocking the recruitment of the viral replication complex and the RNA polymerase II (Pol II) transcription complex at OriLyt. It is presently unknown how this is occurring, but it might suggest that unknown restriction factors at OriLyt are targeted by CRLs leading to their polyubiquitylation via K48-linked ubiquitin (thus targeting them for degradation) or K63-linked ubiquitin suggesting a CRL-mediated signaling requirement during lytic reactivation. Modification (either Ub or NEDD8) may also be required for the proper processing of proteins that are essential for reactivation. Nevertheless, this study has highlighted that NEDDylation may be a novel target for the treatment of various KSHV-associated malignancies.